4 ms·
As a roboticist, this video made me very angry. This is not a great showing of science fact, but a strange distortion of the current state of robotics to fit a
by ModernMech 12y ago
As a roboticist, this video made me very angry. This is not a great showing of science fact, but a strange distortion of the current state of robotics to fit a particular narrative. By the end, it definitely ventures well into the realm of science fiction. I'll point out some of the most egregious examples. This is a pretty long post, but if you spent 15 minutes watching the video, the least you can do is spend 15 minutes to read the following counterpoint.
Self-service checkout – “What used to be 30 humans is now 1 human overseeing 30 self-service check outs” – At all the grocery stores I’ve been to in my area, there are still ~30 human-staffed checkout lanes and 4 – 6 self-service machines, which replaced maybe two rows of check outs and still employs a human for oversight. However, these self-service checkouts allow the stores to operate 24/7, because after 10:00 they are the only lanes open. Now the store is open longer so it can make more money, whereas before they closed at 10:00. More money for the store means perhaps a whole new store opening, employing more people than would have otherwise been employed without self-service lanes.
Self-driving cars – “They are not the future, they are here and they work” – For narrow definitions of "here" and "work". There's a big difference between being developed in a lab and driving up and down the coast in sunny Cali, and being deployed to millions of consumers around the globe in multitudinous climates. He lists some things self-driving cars do better than humans, but fails to mention that self-driving cars lack the complex decision making, cognition, and communication abilities that are required in every-day driving. Self-driving cars are low hanging fruit for roboticists because we have a set of rules governing transportation, and that’s prime material for automation. Take the rules, code them up, done. Problem is, the rules are broken constantly, and in a way that driving a million miles up and down the west coast won’t solve (How many miles exactly has the Google car driven in snow?). Self-driving cars will come, but as long as there are still human drivers and weather, they will not live up to the hype they are currently generating. Take it from someone who actually engineers these types of robots.
Law Bots – He claims that bots will replace the bulk of lawyering because they can sift through a million e-mails quickly. And what happens when the bot is done? The lawyer gets back to work. It sounds to me like the bot is making the lawyer more efficient. Instead of spending weeks going through e-mails, he can use a computer to help him, and do his job more efficiently. So maybe we need fewer lawyers? The only reason we needed more in the first place was because of the amount of e-mail and electronic documents created by computers. To say we shouldn’t use a computer to fix a problem caused by computers is beyond ridiculous.
Doctor bots – He says that knowing the reaction of every drug between every other drug is beyond the scope of human capability… sounds like a great job for robots then! Again, this is an instance of a robot augmenting the capabilities of a human, not replacing one. He goes on to say that the doctor bots will be able to read the latest research, understand it, and then go on to implement the knowledge. First, one has to wonder how the research is being generated if human doctors are being replaced by robots. Second, as long as we’re assuming an AI advanced enough to read a research paper, generate inferences and implications based on that research, and act on that new knowledge, we might as well assume that all disease has also been cured, eliminating the need for doctor bots in the first place.
Creative bots – Finally, and most egregiously. Creativity is right up there next to sentience in the totem pole of AI capabilities, and is well into the realm of science fiction. The (only) example he gives, Emily Howell as a source of machine creative music is a complete joke. Here’s what the inventor of the bot has to say about it:
> Professor Cope argues that Emily Howell’s music is still predominantly created by humans. “Computers are not separate things,” he said. “The computer is human-made. The program itself is human-made. The music in the database is human as well. There’s so much about this that is human. There’s just a lot more humans involved in making this than usual.”
Emily Howell is a program designed by a human to output a specific kind of music. It may be music that has not been written before by any human, but it is not creative. In the end, the music is the result of an algorithm, designed by a human. In my view, a creative machine is one that performs a task, on purpose, it was not programmed to perform.
So those are the big problems I have with what he showed. What is even more aggravating is what he left out. What about robots that do things humans can't do. What about a robot that can fit into a 1/2" pipe, or one that can fly (we call them drones), or one that can go into a nuclear reactor or the bottom of the sea. Those robots are taking away 0 jobs, and create hundreds (if not thousands) of jobs for those that design, engineer, manufacture, repair, sell, program, manage, and use those robots.
Finally, robots are great at replacing some tasks that can be automated, but not so great at others. We've had robot receptionists for a while now. They are so terrible, that having a human receptionist is now a competitive advantage for some businesses. I think the general point that this video misses, is that robots are terrible at interfacing with people. Any human-facing occupation, even if it has aspects that can be automated, will probably still exist no matter how pervasive robots become in our lives. When robots can interact with humans well, we're probably talking about a future that is too far out for us to accurately predict.
- dfabulich 12y ago> To say we shouldn’t use a computer to fix a problem caused by computers is beyond ridiculous. I think you may have misunderstood the purpose of the video. The video doesn't say, "hey, we need to turn off these computers because they'll make us obsolete." He's arguing that we'll inevitably use computers to solve a problem created by computers, and that a lot of people will be unemployable as a result, and that's something we need to think about as a society. The video doesn't suggest any concrete solutions at all. (Basic income/negative taxation seem like obvious candidates; walking back technology itself is probably not a good candidate, though some people will certainly try.)
- Houshalter 12y agoHe's not saying that robots are going to take literally every job ever. Only the majority of them. I believe I read that stores don't encourage self-checkouts very much because having humans around significantly decreases shoplifting. This may be less necessary if we get better AIs for monitoring security cameras. I've never read anything about self-driving cars being impossible in bad weather, only that no has done it yet. Machine vision systems are not as vulnerable to this btw and that has been improving in leaps and bounds over just the last 10 years. Lawyers still searched through documents tediously before computers. It was probably even harder then, not easier. It's at least theoretically possible create a robot that can diagnose and prescribe treatments for a patient just like a doctor. All the video claimed is that they can do statistics and search the relevant research a lot better. Which is definitely true, and see IBM's Watson. Arguments about what "creativity" means are never useful or productive. But AI that generates music well already exists and is going to continuously get better.
- zecho 12y agoI don't consider a self-checkout lane to be a "robot" in any sense. It's a customer doing the job of a once trained employee. If you've ever been behind someone new to one of these things, they're slow and annoying experiences. The main reason for them is reducing workers necessary during slow sales periods (like overnights, or weekday mornings). And anyway, the job wasn't replaced by a robot in this case. The job was replaced by a customer.